Draft for ISSS-ISAC Please do not quote or cite without author s permission

Transcription

1 1 After the End of History, Before Everything Changed: Civil Wars, Democratization, and the End of the Cold War Adam Knight Doctoral Candidate Rutgers University New Brunswick The of the Cold War brought an to a state of affairs in which every armed conflict, no matter how small, existed within the context of a larger, comprehensive, bipolar competition. This competition s cessation coincides with a momentary increase in the number of civil wars which by way of negotiation. This suggests that civil war termination underwent a sea change in The civil war termination literature has generally acknowledged this development as well as the fact that it crested about ten years after but for the most part not examined its impact. In accounting for this paradigm shift, I demonstrate that civil wars which ed between 1991 and 2001 and did so by way of a negotiated settlement are not only more likely to experience movement towards democracy, but that this democratizing moment comes more quickly than in the case of victories or negotiated truces. On the other hand, negotiated settlements have a much smaller impact in wars which ed either before 1991 or after This highlights some of the contextual factors be they superpower competition or post-9/11 American foreign policy which may interact with the much-debated impact of different means of civil war termination. Draft for ISSS-ISAC Please do not quote or cite without author s permission

2 2 Background For all the sound and fury regarding how civil wars ought to, it appears to have signified little or less than one would hope given how much time political scientists have spent at any rate. There have been enough legitimate attempts by reputable scholars to answer the question that most of the different possible methods of ing a conflict have a scholarly champion supporting it. This paper does not purport to the debate, but it does seek to clarify it. Specifically, it seeks to provide a more nuanced depiction of the impact of outcomes as well as introducing a temporal element the existence of which few would deny, but fewer have actually studied. The debate over the ideal method of civil war cessation has measured this ideal by the resulting peace s ability to do two things: remain in place as long as possible and encourage democratization. 1 While the latter is still a relatively novel subject of research, former has been subject to more analysis. For instance, Edward Luttwak closed the 1990 s with a sobering observation: perhaps the post-cold War efforts to wars prematurely have caused more trouble than if these conflicts had been left alone. 2 He argued that leaving wars unresolved leaves open the possibility for recidivism. If belligerents have the ability to pick up their arms and fight again, they may well decide to do so, especially if the raison de guerre remains unchanged. Therefore, if wars are allowed to reach their logical conclusion, they are less likely to recur. As shocking as it must have seemed, the notion did have support in the literature. 3 For instance, Harvey Waterman echoes the sentiment that the prosecution of a war provides necessary information for striking a deal to secure its, whatever that may be. 4 Moreover, he notes that absent the information a war provides, a deal may not be possible. While Luttwak s thesis may not have been altogether novel, it sent shockwaves through the discipline, inspiring the scholars to invest time and effort into the connection between outcome and the longevity of the resultant peace. Luttwak s thesis informed Monica Duffy Toft s work on the subject, which also argues that victories can indeed lead to more resilient peaces. She argues that negotiated settlements may not be ideal because they are so difficult to effectively execute as negotiated s to conflicts do not necessarily provide resolution to the aggressor s casus belli, nor do they remove the ability of either side to reinitiate hostilities. 5 Because of this difficulty and the usually insufficient political will to stick to a good settlement 6 many such settlements are doomed to 1 Roy Licklider s piece introduced the first goal but also mentioned a third goal preventing large-scale reprisals which unlike the other two (peace, democratization) has remained relatively lightly examined (see: Roy Licklider, The Consequences of s in Civil Wars, , The American Political Science Review, Vol. 89, No. 3 (Sep., 1995), pp ). 2 Edward N. Luttwak, Give War a Chance, Foreign Affairs, Vol. 78, No. 4 (July/Aug. 1999), pp See: Licklider, Roy, The Consequences of s in Civil Wars, , The American Political Science Review, Vol. 89, No. 3 (Sep., 1995), pp Harvey Waterman, Political Order and the of Civil Wars, Stopping the Killing: How Civil Wars End, ed. Roy Licklider, New York University Press, 1993, pp Monica Duffy Toft, Securing the Peace: The Durable of Civil Wars, Princeton University Press, 2010, pp. pp Kai Frithjof Brand-Jacobsen and Carl G. Jacobsen, Beyond Mediation: Towards More Holistic Approached to Peace-building and Peace Actor Empowerment, Searching for Peace: The Road to TRANSCEND, eds. Johan Galtung, Carl G. Jacobsen, and Kai Frithjof Brand-Jacobsen, Pluto Press, 2002, pp

3 3 fail and the wars they seek to are destined to resume. Military victories remove the ability of one side or another to resume the conflict, thus removing the incentive for either side to restart the conflict; the victorious side can operate without fear of their defeated opponent and the defeated party lacks the means to effectively resist. However, her thesis departs from Luttwak in two ways. First, she notes that victories may also be more conducive to democratization than negotiated s to civil wars. Second, she notes that not all victories have similar impacts on postwar outcomes it matters who wins. 7 Toft points out that negotiated settlements do provide one element critical to a lengthy peace which she refers to as help that is, the positive incentive to cooperate. 8 The problem with negotiated settlements is that they lack the second important element which she refers to as harm that is the disincentive to defect in the form of effective enforcement. All military victories provide harm sufficiently to secure a lasting peace. However, only rebel victories provide the help necessary to allow for democratization. Victorious incumbents, she argues, are driven to punish their defeated adversaries, meaning that as soon as these adversaries are able, they will likely defect against their oppressors. On the other hand, victorious rebels may feel their legitimacy is more tenuous than victorious incumbents, meaning that they are more likely to provide incentives to cooperate in the form of liberalization. While Toft stops short of arguing that this liberalization is necessarily democratization, her use of Polity IV data suggests the terms can be used interchangeably in this context. Toft s theory has attracted considerable criticism as well as plaudits, though. While much of the debate regarding her findings has focused on her conclusions regarding the survivability of peaces secured by various outcomes, some of this criticism has also addressed these outcome s supposed impacts on democratization. Indeed, Marcus Nilsson s examination of the very same question produces nearly opposite results. He finds that negotiated settlements are substantially more conducive to political contestation and inclusion (his Dahlian proxies for democracy) in the postwar context. 9 On a similar note, Laurie Nathan expressed concern that Toft s results may lead Western powers to abandon efforts to wars before one side has defeated the other. 10 This is especially important given that small tweaks in how cases are coded substantially change Toft s results. 11 A more conceptually rooted challenge can be found in Leonard Wantchekon s theoretical examination of postwar democratization. He proposed a method of democratization linked with the manner in which a civil war s in which the combatants perceive the gains to be made by such a change. 12 This can only take place after a negotiated settlement, because otherwise the parties would lack the mutual leverage necessary to guarantee the establishment of Schumpeterian electoral instruments. He deliberately goes no further than arguing that this method may take place, as he posits that the resulting government may not necessarily mature 7 Toft, Securing the Peace, Ibid. pp Marcus Nilsson, Reaping what was sown: Conflict outcome and post-civil war democratization, Cooperation and Conflict, Vol. 47, No. 3 (September, 2012), pp Laurie Nathan, and Monica Duffy Toft, Correspondence: Civil War s and the Prospects for Peace, International Security, Vol. 36, No. 1 (Summer, 2011), pp Ibid. 12 Leonard Wachekon, The Paradox of Warlord Democracy: A Theoretical Investigation, The American Political Science Review, Vol. 98, No. 1 (Feb., 2004), pp

4 4 into a consolidated democracy, but holds on to the notion that this does not necessarily doom the democratized government to collapse. While his theory is not accompanied by statistical analysis, it poses a serious conceptual challenge. Democratization In light of this, it may be worthwhile to re-examine the possibility that negotiated settlements may better produce postwar democratization. Nilsson s argument is pertinent here, but perhaps Wantchekon s depiction is more useful in describing exactly how it might come about. Wantchekon s Warlord democratization is one which is intentionally limited. 13 In order for elites of the different sides to actually prefer to democratize, they themselves must perceive some sort of incentive on their part relative to their other choices. 14 Those incentives come from the particular limitations on contestation. By limiting the degree of contestation, Wantchekon s warlords-cum-governors may be able to ensure that they can maintain a certain degree of control over their own political destiny, a development which many political scientists have identified as important for an embryonic postwar democracy. For instance, Caroline Hartzell and Matthew Hoddie s defense of powersharing arrangements posit a mode of democratization where contestation is intentionally limited. 15 This is done to assuage fears of domination and in so doing encourages the conflict to come to an. Moreover, Hartzell and Hoddie s defense of power-sharing agreements bears something of a resemblance to Ar Lijphart s defense of consociational democracy. 16 Lijphart defs the mode of democracy as a means of providing some degree of stability to democracies in deeply fragmented societies. 17 While this development is distinct from creating an entirely new democratic regime, several of the elements in Lijphart s defense resemble Hartzell and Hoddie s description. For instance, one of the most critical elements necessary for consociationalism is that elites recognize the dangers of the system s collapse. Wantchekon appears to paraphrase the idea behind this in his description of the game warlords play in assenting to democratization: they see settling and coexisting with one s adversaries as preferable to either surrering to that adversary or continuing to prosecute the war. This is not to say that the fit is ideal, though. Hartzell and Hoddie emphasize the importance of securing elite cooperation by assuaging their fears, but these fears are not quite the same as those discussed in Lijphart. The elites in Lijphart s conception are not (necessarily operating under conditions of potential (or actual) civil war, for one. Lijphart argues that in the absence of cross-cutting cleavages within a society s masses, the society s elites must make pains to cooperate especially with rival groups. 18 This is an especially difficult task if one assumes that elites may not be prepared to take such pains, as are likely the case under conditions of a civil war. Regardless, that society s elites must find ways to cooperate across rival groups or else the system is in danger of collapse. 13 Wantchekon, Warlord Democracy, pp Those being continuing the conflict and, in Wantchekon s parlance, inviting leviathan, which serves as a representation of surrer. 15 Matthew Hoddie, and Caroline A. Hartzell, The Art of the Possible: Power Sharing and Post-Civil War Democracy, World Politics, Vol. 67, No. 1 (Jan, 2015), pp In part because the critics of power-sharing cite similar issues as did critics of consociational democracy 17 Ar Lijphart, Consociational Democracy, World Politics, Vol. 21, No. 2 (Jan. 1969), pp Ibid. p. 216

5 5 Power sharing as a method of securing democracy from negotiated settlement is not without critics, 19 but it is feasible (or at least possible). An alternative to this particular method can be found Philip Roeder s description of power dividing. 20 This method diminishes the degree to which the different govern themselves, which he argues increases the need to cooperate and generates cross-cutting cleavages. Regardless of power sharing s merits or the supposed superiority of that of the less consociational power dividing, both require a negotiated settlement to have taken place to be employed. Timing is Everything In order for this sort of arrangement to occur, a negotiated settlement must take place. As such it is necessary for negotiated settlement as a means of ing the war to be considered feasible. As such, an examination of the impact of negotiated settlements should be limited to the post-cold War era. During this era, negotiated settlements became much more common; before the Cold War s close they were possible, but quite rare. 21 Although they remain more common than they were during the Cold War, contemporary negotiated settlements are less likely than during their boom in the 1990 s. Although multiple scholars have noted that negotiated settlements have only become likely outcomes since the Cold War ed, the fact that the impact of this threshold has not been examined previously represents an unfortunate tency in our discipline towards ahistoricity in analysis, particularly in large-n studies. 22 This is unfortunate, as the Cold War represents a substantial sea change in international politics. Every armed conflict had been viewed through the lens of a greater bipolar geopolitical conflict between the United States and the Soviet Union. Thereafter, conflicts could be seen as predominantly concerning those doing the fighting and those who live in its direct proximity. This meant that the international community (including the ascant American superpower) could dedicate itself to the project of stopping wars rather than making sure the right side won. 23 The resultant democratization events gave credence to the notion that liberal democracy would continue its spread unabated. 24 This was unfortunately short-lived. Shortly after Luttwak suggested that negotiated settlements were not conducive to a resilient peace, the terrorist attacks of September 11, 2001 dramatically shifted the US s foreign policy aga. Johan Galtung argues that the attacks led to American leadership to revert to something of a Cold War mindset conflicts, no matter how 19 Donald Rothchild, and Philip G. Roeder, Power Sharing as an Impediment to Peace and Democracy, Sustainable Peace: Power and Democracy after Civil Wars, eds. Philip G. Roeder and Donald Rothchild, Cornell University Press, 2005, pp Philip G. Roeder, Power Dividing as an Alternative to Ethnic Power Sharing, Sustainable Peace: Power and Democracy after Civil Wars, eds. Philip G. Roeder and Donald Rothchild, Cornell University Press, 2005, pp Toft notes the post-cold War increase in settlements (Toft, Securing the Peace, p. 7). 22 Michael Parenti, Patricians, Professionals, and Political Science, The American Political Science Review, Vol. 100, No. 4 (Nov., 2006), p. 502; this is by no means an indictment of political science writ large. Indeed, many political scientists have produced work that is deeply mindful of the context in which the empirical data exists, including the temporal dimension of that context. 23 Marie Olson Lounsbery and Frederic Pearson, Civil Wars: Internal Struggles, Global Consequences, University of Toronto Press, 2009, pp Francis Fukuyama, The End of History? The National Interest, No. 16 (Summer, 1989), pp. 3-18

6 6 remote, must be seen through the lens of how it impacts the war on terrorism. 25 This has led to a more antagonistic perception of rebel movements and diminished the political will to mediate conflicts. This also may have led to a more fragile state of affairs for conflicts which in a negotiated settlement, as the outside political capital which may have been more common during the 1990 s became scarcer thereafter. Thus, negotiated settlements may be less resilient. This should not only make recidivism more common, but it should also make democratization rarer as the United States and its allies have, in general, favored democratization of non-democratic states emerging from civil wars in the wake of the Cold War. Although it still supports democratization in general, the methods with which it does the supporting have changed in the wake of 9/11. Hypothesis While changes in the relative rarity of an event over time may not necessarily have an impact on the particulars of that event, the theory I have posited here suggests that this is the case for negotiated settlements. It appears that conditions may have been ideal for post-negotiated settlement democratization between the of the Cold War and 9/11. As such, I am reexamining Nilsson s hypothesis regarding negotiated settlements, as he argues that they are more conducive to postwar democratization than other outcomes. H1: Countries are more likely to experience a post-civil war democratization if that civil war ed by way of a negotiated settlement. Of more interest is the particular impact of the different time periods, though. This ideal period in which negotiated settlements were common as it was after the of the Cold War and the international community featured powerful states which could mediate without fear of greater geo-political or national security consequences before 9/11 should also be especially conducive to post-negotiated settlement democratization. H1A: Countries are more likely to experience a post-civil war democratization if that civil war ed by way of a negotiated settlement, assuming the conflict in question ed between the of the Cold War and September 11, The design of this paper also allows me to reexamine the Toft hypothesis. It may be the case that the coding differences between Toft and Hartzell s data cause the designs to paint a different picture of the data. It is also possible that the changes I have made to the original design may do the same. As such, I include the Toft hypothesis as another hypothesis for the purposes of the paper. H2: Countries are more likely to experience a post-civil war democratization if that civil war ed by way of a rebel military victory. I expect that rebel military victories should react similarly to negotiated settlements during the 1990 s. It stands to reason that a less antagonistic and more democratization-oriented foreign policy from the United States would have a similar impact on rebel victors as combatants who have agreed to put aside bloody competition. As such, the particular impact of rebel 25 Johan Galtung, 11 September 2001: Diagnosis, Prognosis, Therapy Searching for Peace: The Road to TRANSCEND, eds. Johan Galtung, Carl G. Jacobsen, and Kai Frithjof Brand-Jacobsen, Pluto Press, 2002, pp

7 7 victories during the decade in between the of the Cold War and 9/11 shall be examined as well. H2A: Countries are more likely to experience a post-civil war democratization if that civil war ed by way of a rebel military victory, assuming the conflict in question ed between the of the Cold War and September 11, Design These hypotheses are tested with two methods here. The first is a set of regressions and the second is with a pair of Cox proportional hazard models. In many ways, the regressions mirror Toft s. For instance, outcomes are broken up in much the same way. Victories are broken up according to the victor incumbent (the state) and rebels. cessations are also divided into two categories based on whether the agreement is indefinite. If the agreement states that the remaining sides shall coexist indefinitely, the war is treated as having ed by way of negotiated settlement. Otherwise, a war with a negotiated will be considered to have ed by way of a negotiated truce. Although both designs rely on the venerable Polity IV dataset, 26 this design includes an important innovation of Toft s method of measuring the data from the set. Whereas Toft measures polity scores at five year intervals, 27 this design takes the average scores between the intervals. This may appear a small configuration, but it can have a tremous impact on the results. One reason for this is that it more completely captures the variance in democracy and autocracy leading up to and following a civil war. Another reason for this is that under certain circumstances (for instance, foreign occupation), there may not be a figure for a given country at a given year. As a result, for such a country, there is no data at that particular benchmark. With a more complete picture of democracy (and fewer missing values), one can expect different results to emerge. Also similar to Toft s design, the case selection s before the present day. The cases for this design will come from Caroline Hartzell s military integration dataset. 28 Hartzell s data includes all civil wars which began after WWII and which ed by the close of 2006 according to the Correlates of War definition of a civil war. 29 This allows for each case to contribute democratization data for at the very least ten postwar years. Additionally, Hartzell s data contains a number of variables which makes it a natural fit with the design, including outcome data. More importantly, it contains variables which will be especially helpful in constructing a set of control variables. 26 Monty G. Marshall, Ted Robert Gurr, and Kieth Jaggers, Polity IV Project: Political Regime Characteristics and Transitions, , Center for Systematic Peace, (accessed June 20, 2017) 27 Toft, Securing the Peace, pp Caroline A. Hartzell, Mixed Motives?: Explaining the Decision to Integrate Militaries at Civil War s End, New Armies From Old: Merging Competing Military Forces after Civil Wars, eds. Roy Licklider, Georgetown University Press, 2014, pp There are a total of 128 cases included in the dataset.

8 8 Using Virginia Page Fortna s scale for measuring a peacekeeping mission s difficulty as an instructive example, 30 I have selected several additional controls already present in Hartzell s dataset. These controls include variables accounting for whether the war was over identity, the war s intensity, its duration, the number of in the war, the GDP/pc at the war s, and whether or not the resulting peace failed. 31 Fortna uses her scale to measure the degree of difficulty in an analysis of peacekeeping s effectiveness, as she argues that previous examinations suffer from unaccounted for spurious relationships. 32 While she has a direct and apparent theoretical justification for doing so that this design lacks, it nonetheless is appropriate. Even if negotiated settlements are not necessarily more common in especially difficult cases (where either peace or democracy is concerned), one can plainly see the import to controlling for such difficulty in any case. The hazard models examine the timing of the transitions to two of polity IV s four categories of regimes democracy and open anocracy 33 in the postwar setting and the impact that different outcomes have on the timing of these events. A transition assumes a change, so similarly to the regression analysis, the hazard models shall observe the change to democracy or anocracy. This means that a democratic regime must not have been in place at any time during the five years prior to the war s initiation for that regime to be considered to have democratized in the time examined. The same goes for open anocratic regimes. The hazard models also include the same battery of controls as do the regression models. Results Table 1 WWII-2006 End to 5 years 6-10 years years years 2.813** (1.062) 3.541*** (1.215) 3.205** (1.364) 3.366** (1.583) Rebel.228 (1.382) (1.641) (1.674) (1.832) *** (1.123) *** (1.296) ** (1.434) ª (1.605).275 (1.498).414 (1.705).391 (1.944) (2.165) Table 2 WWII-1990 End to 5 years 6-10 years years years (1.690) ª (1.915) ª (2.002) ª (2.181) Rebel.589 (1.609) (1.904) (1.941).478 (2.114) 30 Virginia Page Fortna, Does Peacekeeping Work?: Shaping Belligerents Choices after Civil War, Princeton University Press, 2008, pp , There was an additional control on the Fortna scale for democracy which was not included for reasons which should be obvious. 32 The correlation between fires and eventual presence of firefighters problem. 33 For reference, a regime is considered to be a democracy with a score of 6 or higher, while an open anocratic regime has a score of between 1 and 5.

9 *** (1.263) *** (1.443) (3.035) (3.459) *** (1.571) (3.672) *** (1.738) (3.994) Table End to 5 years 6-10 years years years 4.221*** (1.192) 5.492*** (1.337) 5.611*** (1.442) 6.044*** (1.632) Rebel (2.217) (2.726) (3.156) (3.076).166 (1.991) (2.260) (2.390) (2.584) (1.798) (2.046) (2.176) (2.377) Table End to 5 years 6-10 years.730 (1.770) (2.010) A casual glance at the regression results points to two observations regarding the hypotheses. First, negotiated settlements presage substantial movement towards democratization, but only in the ten years which followed the close of the Cold War. Secondly, rebel military victories are less likely to presage de-democratization than incumbent military victories, but only during the Cold War. However, taken as a whole rebel victories do not appear to have a significantly positive impact on democratization as operationalized here. The last time period the regression analysis covers produces results which require some explanation. This period (September 11, 2001 to the of 2006) is quite short, so it is hardly a surprise that there are fewer civil war terminations during this block of time than the other two. The peculiarities of the sample s size preclude certain analyses, though. There are too few nonnegotiated settlements to trust the regression results validity for any of these three outcomes. Even the negotiated settlement results should be taken with a grain of salt. The relationship is neither particularly substantial nor significant and that this is a noteworthy change from those from the previous decade, but beyond this these results should be treated with a healthy amount of skepticism. Table 5 Democratization WWII-2006 WWII * (1.325).206ª (.217) 3.578** (2.220) Rebel.200* (.172).200* (.172) Omitted

10 10.384** (.188).248** (.138) (1.390) 6.485*** (4.333) Omitted 4.646** (3.096) Table 6 Open Anoc. WWII-2006 WWII (.950).365 (.334) 3.706** (2.344) Rebel.705 (.360).610 (.324) (2.734).279** (.177).189*** (.122) (.799) 6.603** (4.983) Omitted 4.727* (4.037) The hazard models produce two expected results and one profoundly curious one. While the results for civil wars ed by way of rebel victory and negotiated settlement mirror those from the regression analysis, the results for negotiated truces could merit further examination. It appears that truces predict democratic (and anocratic) transition more reliably than any other outcome. There are reasons to be skeptical of the implications of this result, though. For one, there are fewer negotiated truces than any of the other three outcomes in the set, meaning that a few odd cases may cause the entire category to be misrepresented in the results here. Also, the notion that not agreeing to coexist indefinitely with one s adversary seems like a profoundly odd basis to build a democratic (or more democratic) regime. Discussion It may well be that the circumstances around the of the Cold War are not replicable in a meaningful way here. It may be that the of global bipolar conflict combined with an explicitly pro-democratic international community produced a wave of democratization that may never be seen again. However, what can be said is that something about that decade made negotiated settlement a means of terminating a civil war which encouraged democratization. From a strictly theoretical perspective, this analysis highlights the importance of temporal context in large-n conflict studies. Each point of data in a conflict study happens within a rich context that can be difficult to capture effectively in statistical analysis. While this analysis may not necessarily be exemplary, it can at least highlight the importance with which this context should be treated by other scholars. The results seem to call into question Toft s theory on the subject. It also suggests that Galtung may be on to something. 34 It may well be that by viewing conflicts through the lens of the war on terror, the United States has cooled support for conflict mediation and in doing so 34 Galtung, 11 September 2001, pp

11 11 diminished the impact of negotiated settlements. Even if this effect is not to blame, the results suggest that efforts to civil wars before they have resolved themselves may be worthwhile if one hopes for democratization to take place in the war s wake. However, Toft s theory also speaks to the likelihood of recidivism, 35 which this analysis does not. One should therefore take care not to mistake democratization for a resilient peace. There are multiple reasons to be cautious when making policy from these results, however. For one, as James Vreeland astutely observed, the polity set is a bit problematic for civil war analysis. 36 Using his x-polity correction circumvents the conceptual issues, but his reinterpretation is not without its own issues. The x-polity data only measures the institutions and practices surrounding the executive, making the polity set even more conceptually shallow. This is not to say that replicating this design with x-polity data would not be merited, but it does mean that anyone doing so should also be skeptical of those results as well. Another issue lies in the fact that this data also relies on the polity authors polity2 interpolation of missing data. This has led to some curious shifts in polity scores. Thomas Plümber and Eric Neumeyer have proposed some alternatives, which, in conjunction with replication with the x-polity interpretation of the polity data, could make these results more robust. 37 Similarly the results could be improved by replicating them using different data sets entirely. Two in particularly noteworthy ones are the Varieties of Democracy 38 data set and Boix, Miller, and Rosato s dichotomous democracy data set. 39 The design for both methods could be refined further to better represent the relationships between the models variables. This is especially the case with the hazard models. For instance, the hazard model does not account for pre-existing democracy or anocracy, meaning that the output here is somewhat misleading. Dropping democratic (and anocratic) cases as appropriate could help to solve this issue, as could replicating the analysis testing for de-democratization (and de-anocratization). Even this is not without some issues, though. The categories of democracy and open anocracy are somewhat arbitrary. Civil war termination methods may not be a terribly precise method to predict postwar political developments, but this analysis does shed some light on its impact. More to the point, it introduces a contextual element to this effect. While it may be that supporting negotiated settlements as a means of ing a civil war is unwise if one hopes for postwar democratization, this analysis suggests that if this is so it is a new development. It may be that if the international 35 The same can be said for Luttwak s. 36 James Raymond Vreeland, The Effect of Political Regime on Civil War: Unpacking Anocracy, The Journal of Conflict Resolution, Vol. 52, No. 3 (June 2008), pp Thomas Plümber, and Eric Neumayer, The level of democracy during interregnum periods: recoding the polity2 score, Political Analysis, Vol. 18, No. 2 (2010), pp Michael Coppedge, John Gerring, Staffan I. Lindberg, Sv-Erik Skaaning, Jan Teorell, David Altman, Michael Bernhard, M. Steven Fish, Adam Glynn, Allen Hicken, Carl Henrik Knutsen, Kyle L. Marquardt, Kelly McMann, Valeriya Mechkova, Pamela Paxton, Daniel Pemstein, Laura Saxer, Brigitte Seim, Rachel Sigman, and Jeffrey Staton V-Dem Codebook v7.1. Varieties of Democracy (V-Dem) Project, accessed September 13, Carles Boix, Michael Miller, and Sebastian Rosato, A Complete Dataset of Political Regimes, , Comparative Political Studies, Vol. 46, No. 12 (Dec, 2013), pp

12 12 community were to become more invested in ing wars before their logical conclusion, the post-cold War impact of negotiated settlements may emerge once again. Appix This appix contains a more complete version of the results. These versions are arranged slightly differently to accommodate the inclusion of the control variables. Regressions (WWII-2006) End to 5 years *** (1.062) 3.541*** (1.215) 3.205** (1.364) 3.366** (1.583) Stake * (1.009) (1.160) (1.275) (1.438) lnwardur.620 (.318).549ª (.361).473 (.401) (.444) Lnintensity (.266) (.301).013 (.359).004 (.409) (.343) (.411) (.389) (.429) *.0007** Peacefail (.987) (1.124).422 (1.222) (1.375) Constant.681 (2.661) (3.012) (3.462) (4.033) (WWII-2006) End to 5 years Rebel.228 (1.382) (1.641) (1.674) (1.832) Stake (1.067) ª (1.232) * (1.337) (1.506) lnwardur.935*** (.322).859** (.368).774* (.397).293 (.438) Lnintensity (.282) (.322) (.376) (.428) (.359) (.428) (.398) (.437) *.0007** Peacefail (1.017) (1.169).164 (1.254) (1.414) Constant (2.724) (3.104) (3.530) (4.137)

13 13 (WWII-2006) End to 5 years *** (1.123) *** (1.296) ** (1.434) ª (1.605) Stake (1.008) (1.179) (1.304) (1.483) lnwardur.708** (.300).720** (.346).679* (.380).196 (.421) Lnintensity (.262) (.303) (.359) (.413) (.338) (.412) (.390) (.432).0004* Peacefail (.968) (1.126).116 (1.226) (1.391) Constant (2.597) (3.001) (3.485) (4.136) (WWII-2006) End to 5 years (1.498).414 (1.705).391 (1.944) (2.165) Stake (1.054) ª (1.218) * (1.321) (1.479) lnwardur.914*** (.309).921*** (.353).835** (.382).321 (.421) Lnintensity (.284) (.324) (.374) (.426) (.359) (.432) (.402) (.440) *.0007** Peacefail (1.036) (1.191).141 (1.278) (1.427) Constant (2.828) (3.229) (3.611) (4.218) (WWII-1990) End to 5 years (1.690) ª (1.915) ª (2.002) ª (2.181) Stake ª (1.030) ª (1.182) * (1.282) (1.441) lnwardur.943*** (.308).975*** (.348).893** (.377).378 (.418)

14 14 Lnintensity (.273) (.310) (.354) (.423) *.0007** Peacefail (1.013) (1.157) Constant (2.716) (3.079) (.364) (.400).220 (1.237) (3.526).013 (.416) (.439) (1.393) (4.117) (WWII-1990) End to 5 years Rebel.589 (1.609) (1.904) (1.941).478 (2.114) Stake (1.062) ª (1.219) * (1.343) (1.512) lnwardur.959*** (.326).889** (.375).811** (.403).348 (.443) Lnintensity (.277) (.316) (.374) (.425) (.355) (.429) (.398) (.436).0007**.0006* Peacefail (1.016) (1.171).183 (1.255) (1.413) Constant (2.721) (3.111) (3.535) (4.139) (WWII-1990) End to 5 years *** (1.263) *** (1.443) *** (1.571) *** (1.738) Stake (.977) (1.133) (1.266) (1.429) lnwardur.631** (.294).609* (.336).590ª (.372).084 (.409) Lnintensity (.255) (.292) (.350) (.400) (.329) (.397) (.378) (.415) Peacefail (.942) (1.086).056 (1.195) (1.347) Constant ª

15 15 (2.539) (2.906) (3.391) (3.992) (WWII-1990) End to 5 years (3.035) (3.459) (3.672) (3.994) Stake (1.045) (1.210) * (1.321) (1.485) lnwardur.902*** (.308).917** (.352).831** (.382).306 (.422) Lnintensity (.274) (.313) (.369) (.421) (.354) (.427) (.398) (.437) *.0007** Peacefail (1.019) (1.173).218 (1.260) (1.417) Constant (2.719) (3.111) (3.549) (4.151) ( ) End to 5 years *** (1.192) 5.492*** (1.337) 5.611*** (1.442) 6.044*** (1.632) Stake (.981) (1.112) (1.212) (1.366) lnwardur.623** (.302).540ª (.337).378 (.372) (.415) Lnintensity (.258) (.289).001 (.340).139 (.391) (.334) (.394) (.367) (.403) *.0007*** ** Peacefail (.957) (1.079).118 (1.161) (1.307) Constant (2.566) (2.870) (3.279) (3.878) ( ) End to 5 years Rebel (2.217) (2.726) (3.156) (3.076) Stake (1.037) ª (1.205) * (1.299) ª (1.457) lnwardur.919***.929***.837**.316

16 16 (.308) (.350) (.380) (.420) Lnintensity (.277) (.319) (.370) (.421) (.359) (.427) (.397) (.435).0006* Peacefail (1.016) (1.167).104 (1.255) (1.410) Constant (2.742) (3.120) (3.525) (4.129) ( ) End to 5 years (1.991) (2.260) (2.390) (2.584) Stake (1.035) ª (1.194) * (1.297) ª (1.447) lnwardur.921*** (.308).937*** (.350).842** (.380).316 (.416) Lnintensity (.274) (.313) (.366) (.415) (.356) (.427) (.398) (.434) *.0006* Peacefail (1.017) (1.167).182 (1.250) (1.396) Constant (2.722) (3.102) (3.522) (4.096) ( ) End to 5 years (1.798) (2.046) (2.176) (2.377) Stake (1.034) ª (1.197) * (1.302) ª (1.463) lnwardur.922*** (.308).931*** (.351).840** (.382).322 (.422) Lnintensity (.275) (.314) (.370) (.424) (.355) (.428) (.399) (.439) *.0007** Peacefail (1.022) (1.177).208 (1.263) (1.420)

17 17 Constant (2.736) (3.126) (3.572) (4.194) ( ) End to 5 years (1.770) (2.010) Stake (1.033) (1.192) lnwardur.904*** (.310).896 (.353) Lnintensity (.276) (.315) (.354) (.426) *.0007** Peacefail (1.048) (1.202) Constant (2.794) (3.184) Hazard Models Democratization (WWII-2006) Rebel Outcome 2.544*.200* (1.325) (.172) Stake (.517) (.243) Lnwardur (.117) (.121) Lnintensity (.123) (.124) (.124) (.136) Peacefail.545ª.709 (.224) (.274).384** (.188) (.509) (.127).873 (.127).992 (.122).537ª (.220) 6.485*** (4.333).534ª (.231) (.113).971 (.148) (.140).675 (.258) Democratization (WWII-1990) Rebel Outcome.206ª.200*.248** Omitted

18 18 (.217) (.172) (.138) Stake.791 (.329).529 (.243).994 (.428) Lnwardur ª (.122) (.121) (.142) Lnintensity (.146) (.124) (.123) (.154) (.136) (.123) Peacefail ª (.274) (.274) (.213) Democratization ( ) Rebel Outcome 3.578** Omitted (2.220) (1.390) Stake (.626) (.330) (.298) Lnwardur (.115) (.114) (.121) Lnintensity (.126) (.126) (.125) (.127) (.127) (.130) Peacefail.513* (.208) (.264) (.286).764 (.329) (.114).863 (.126) (.127).685 (.264) 4.646** (3.096).610 (.261) (.116).914 (.140) (.138).718 (.276) Anocracy (WWII-2006) Outcome (.950) Stake.538ª (.230) Lnwardur (.146) Lnintensity.836 (.108).950 (.163) Rebel.705 (.360).435* (.210) 1.196ª (.143).890 (.118).923 (.160).279** (.177) (.524) (.134).747** (.107) (.167) * 6.603** (4.983).521ª (.220) 1.252* (.153).896 (.114).915 (.164)

19 19 Peacefail (.530).978 (.473) (.601).786 (.383) Anocracy (WWII-1990) Rebel Outcome (.334) (.324) Stake * (.257) (.199) Lnwardur 1.262* (.156) (.142) Lnintensity (.130) (.120) (.155) (.161) Peacefail (.442) (.481) Anocracy Rebel ( ) Outcome 3.706** (2.344) (2.734) Stake ª (.340) (.226) Lnwardur (.138) (.145) Lnintensity (.110) (.112) (.153) (.161) Peacefail (.501) (.529).189*** (.122) (.500) (.139).733** (.100) (.175) * (.714) (.799).511ª (.213) 1.218ª (.147).873 (.112).933 (.159) (.494) Omitted.514ª (.215) 1.208ª.867 (.111).931 (.158) (.478) 4.727* (4.037).530ª (.223) 1.243* (.150).869 (.113).916 (.163).882 (.421)